| The problem of non-matched filter signal detection, identification and/or characterization is of significant interest to military planners who desire increased situational awareness of Radio Frequency (RF) systems operating within in a given area. Situational awareness also provides a management tool for interference avoidance by allowing the integration of new, RF-based, devices without affecting the performance of existing ones. This research introduces an entropy-based spectral processing technique for passively identifying and characterizing communication signals. The proposed technique is based on well-established concepts of sequence entropy or concentration occurring within a specified transformation space of the signal of interest. As demonstrated here, performance of the entropy-based spectral processing technique is dictated by input variables which partition the signal interest, transform the partitioned signal to the spectral domain, and calculate an entropy-based metric for each transformed partition. The process produces the Spectral-Entropy response for the signal of interest. A proof-of-concept demonstration was conducted by applying the proposed technique to both a simulated and an experimentally collected OFDM-Based IEEE 802.11a waveform. Features which emerge within the Spectral-Entropy response are visually correlated with known deterministic and random components of the 802.11a waveform at different Signal-to-Noise Ratios (SNR). These components are readily identified by their comparatively low Spectral-Entropy (high concentration) at SNR's as low as -5 dB. |